Extreme Wind Speeds Software: Fortran Code for Daily Maximum Wind Speeds Data Sets

We provide a pre-compiled executable for Windows 2000/XP platforms. You can compile this code yourself for other platforms if you have a Fortran-77/90/95 compiler available.

If you have a Fortran-77/90/95 compiler, you can modify this code with minimal effort to analyze other data sets.

1. xxx.out

   This file contains a table for each value of the threshold parameter. This table contains:

   • The DeHaan estimates for the shape and scale parameters of the generalized Pareto distribution.

   • The wind speeds corresponding to a set of specified mean return intervals.

   • The maximum wind speed for the fitted distribution.

   • Load factors corresponding to each of the mean return intervals.

2. xxxx.ldf

   For each value of the threshold, the following values are written to the file:

   • the threshold value
   • the number of values above the threshold
   • the estimate of the shape parameter
   • the estimate of the standard deviation of the shape parameter
   • the wind speeds corresponding to each of the return intervals
   • the maximum possible wind speed
   • the load factors corresponding to a return return interval of 50 years

   This file is provided to make it convenient to read the values into another program for further analysis (e.g., a program for plotting the data).

For the sake of discussion, let us assume that you have created the directory "C:\WINDDHN" and that you have copied the data file "abilene.tx" (which contains the data for Abilene, TX) and the program executable "WINDDHN.EXE" to this directory.

To run the program:

1. Open a command prompt window.

2. Enter
   WINDDHN.EXE

3. When you are prompted for the input file name, enter
   ABILENE.TX

4. When you are prompted for the size of the interval, enter 4.

   Since these files contain daily data, there may be significant autocorrelation (and therefore, the input data are not independent) in the data since days close to each other can result from the same storm. To minimize this autocorrelation, the data is split into intervals of x days where x is typically set to either 4 or 8 (the program will accept values for the interval size from 1 to 16). The revised data then consists of the maximum wind speeds of each of the intervals.

5. The program then prompts you for the desired mean return intervals. If you enter either a 0 or a -1, default pre-selected choices will be used. If any other value is entered, the program will prompt for the specific values (up to 10 return intervals can be specified). If a return value of 50 years is not entered, it will automatically be added since this is used to calculate the load factor.

1. abilene.ldf
2. abilene.out

These files can be viewed using Notepad (or any other suitable editor).

For non-Windows platforms, the instructions are similar. Just open a terminal window, initiate the program, and enter the desired data file, interval size, and return values.

• winddhn.f source file
• winddhn.exe executable file for the Windows 2000/XP operating systems.

Note that we have modified the code from the version that was previously available on the NIST ftp site. The output generated from the program is the same. However, the program was rewritten to be easier to adapt to other data sets.

The program contains the following subroutines.

• winddhn

  This is the driver program. Although you should not need to modify the basic structure of this routine, you may want to tweak a few of the PARAMETER statements to accomodate your own data sets. In addition, you can set the desired number of iterations for the threshold (the program uses 25).

• getdat

  This routine reads the input data file. This routine will need to be modified if you are adapting this code to your own data sets.

• dehaan

  This routine generates the DeHaan estimates.

• openfi

  This routine opens the input/output files. This routine also prompts for the desired interval size and the desired mean return intervals.

• intrvl

  This subroutine splits the data into intervals and then extracts the maximum of each interval to form the revised data set.

• maxptr

  This is a utility subroutine used by INTRVL.

• sort

  This subroutine performs a sort of the data using the QuickSort algorithm. Although you can replace this with your own sort routine, this is an excellent sort algorithm and there is little to be gained by replacing it.

1. Modify the GETDAT routine to read your own data files.
2. Check the WINDDHN routine to set the desired threshold iterations.
3. Compile the code on your system.